In this study, Extractive fermentation (ExFerm) was utilized for the simultaneous production and extraction of therapeutic protease using Bacillus subtilis, with agro-industrial waste as the substrate and deep eutectic solvents (DES) as the extraction medium. Betaine: Butylene glycol was utilized as the extraction medium, selected based on the interaction energy of the solvent with the product. The primary objective was to develop a mathematical model, which accurately describes and predicts the performance of ExFerm, and to optimize key phases of the process, specifically the interaction and extraction stages. Non-Dominated Sorting Genetic Algorithm II (NSGA-II) was employed to optimize the process, allowing for the identification of trade-off solutions without compromising the objectives of the process. In this case, the primary objectives were maximizing yield and efficiency while minimizing operational costs. The model encompasses multiple scales of the ExFerm process, identifying optimal Pareto solutions to facilitate informed decision-making for process improvements. The advanced modeling and optimization approach addresses the limitations of traditional trial-and-error methods, ensuring consistent product quality and enhancing the overall efficiency of therapeutic protease production. The findings highlight the essential role of optimization in refining bio-separation processes, paving the way for more sustainable and cost-effective biomanufacturing practices.